October 2025
It’s trivially easy to poison LLMs into spitting out gibberish, says Anthropic
Just 250 malicious training documents can poison a 13B parameter model - that's 0.00016% of a whole dataset Poisoning AI models might be way easier than previously thought if an Anthropic study is anything to go on. …
SonicWall breach hits every cloud backup customer after 5% claim goes up in smoke
Affects users regardless of when their backups were created SonicWall has admitted that all customers who used its cloud backup service to store firewall configuration files were affected by a cybersecurity incident first disclosed in mid-September, walking back earlier assurances that only a small fraction of users were…
Take this rob and shove it! Salesforce issues stern retort to ransomware extort
CRM giant 'will not engage, negotiate with, or pay' the scumbags Salesforce won't pay a ransom demand to criminals who claim to have stolen nearly 1 billion customer records and are threatening to leak the data if the CRM giant doesn't pony up some cash.…
Germany slams brakes on EU’s Chat Control device-scanning snoopfest
Berlin's opposition likely kills off Brussels' bid to scan everyone's messages Germany has committed to oppose the EU's controversial "Chat Control" regulations following huge pressure from multiple activists and major organizations.…
Employees regularly paste company secrets into ChatGPT
Microsoft Copilot, not so much Employees could be opening up to OpenAI in ways that put sensitive data at risk. According to a study by security biz LayerX, a large number of corporate users paste Personally Identifiable Information (PII) or Payment Card Industry (PCI) numbers right into ChatGPT,…
Nearly a year after attack, US medical scanning biz gets clear image of stolen patient data
No fraud monitoring and no apology after miscreants make off with medical, financial data Florida-based Doctors Imaging Group has admitted that the sensitive medical and financial data of 171,862 patients was stolen during the course of a November 2024 cyberattack.…
Police and military radio maker BK Technologies cops to cyber break-in
Florida comms outfit serving cops, firefighters, and the military says hackers pinched some employee data but insists its systems stayed online BK Technologies, the Florida-based maker of mission-critical radios for US police, fire, and defense customers, has confessed to a cyber intrusion that briefly rattled its IT systems…
OpenAI bans suspected Chinese accounts using ChatGPT to plan surveillance
It also banned some suspected Russian accounts trying to create influence campaigns and malware OpenAI has banned ChatGPT accounts believed to be linked to Chinese government entities attempting to use AI models to surveil individuals and social media accounts.…
Britain eyes satellite laser warning system and carrier-launched jet drones
Space sensors and UAVs at sea top MoD's list in new wave of cutting-edge projects The UK is pressing ahead with cutting-edge defense projects, the latest including research to protect satellites from laser attack and a technology demonstrator for a jet-powered drone to operate from Royal Navy carriers.…
UK Home Office opens wallet for £60M automated number plate project
Department eyes new app to tap national ANPR data for live alerts, searches, and integrations The UK's Home Office is inviting tech suppliers to take part in a £60 million "market engagement" for an application that uses data from automated number plate recognition (ANPR) systems.…